Abstract
Gliomas cause high mortality around the world. The metabolic pattern of the tumor was previously suggested to be associated with the patient's survival outcome and immune activity. Yet, this relationship in glioma remains unknown. This study systematically evaluated the immune landscape in different phenotypes classified by metabolic-related pathways of 3068 glioma samples and 33 glioblastoma single-cell sequencing samples. Machine learning prediction analysis of microarray with R (pamr) was used for validating clustering results. A total of 5842 pan-cancer samples were used for external validation of the metabolic clusters. Cell Counting Kit-8 (CCK8) assay, cell clone assay, EdU assay, wound healing assay, Transwell assay, and co-culture assay were performed to verify the distinction in molecular characteristics among metabolic clusters. Metabolomics and RNA sequencing were performed on HS683 and U251 cells to annotate potential hyaluronic acid (HA)-mediated pathways. Three distinct metabolic phenotypes were identified. Metabolic cluster 1 correlated with a high number of immune infiltrating cells and poor survival of glioma patients. Metabolic clusters were proved with different levels of the macrophage markers CD68 and CD163 by multiplex immunofluorescence staining. Glioma cells from other metabolic clusters also expressed various levels of HA. HA was further found to mediate glioma proliferation, progression, and invasion. Moreover, HA potentially promoted macrophage recruitment and M2 polarization through the IL-1/CHI3L1 and TGF-b/CHI3L1 axes. HA also regulated the expression of PD-L1. This work revealed the significant connection between metabolic patterns, especially HA, and tumor immune infiltration in gliomas.